What is RAG? 📚

A practical guide and implementation of Retrieval-Augmented Generation (RAG) - a technique that combines large language models with external knowledge retrieval to provide more accurate, grounded, and contextual responses.

Overview

This project demonstrates how to build RAG systems by:

• Extracting knowledge from documents (PDFs, text files, etc.)
• Processing and structuring that knowledge
• Retrieving relevant information based on queries
• Generating enhanced responses using both retrieved knowledge and LLMs

What is RAG?

RAG is a machine learning approach that enhances large language models (LLMs) by retrieving relevant information from external documents before generating responses. This solves common LLM limitations:

• ✅ Provides current, factual information
• ✅ Reduces hallucinations and misinformation
• ✅ Enables domain-specific knowledge integration
• ✅ Improves transparency by citing sources
• ✅ Works with private/proprietary documents

Project Structure

what-is-rag/
├── README.md                          # This file
├── requirements.txt                   # Project dependencies
├── pyproject.toml                     # Project configuration
├── main.py                            # Main application entry point
└── 1_document_loaders/                # Document loading notebooks
    └── pdf_loader.ipynb               # PDF extraction & parsing examples

Installation

Prerequisites

• Python 3.12+
• pip or your preferred package manager

Setup

1. Clone or download the project

cd what-is-rag

2. Create a virtual environment (recommended)

python -m venv venv
# On Windows:
venv\Scripts\activate
# On macOS/Linux:
source venv/bin/activate

3. Install dependencies

pip install -r requirements.txt

Optional: OCR Support

For advanced PDF image-text extraction with OCR:

Windows:

choco install tesseract-ocr

macOS:

brew install tesseract

Linux:

sudo apt-get install tesseract-ocr

Key Components

1. Document Loaders (1_document_loaders/pdf_loader.ipynb)

Demonstrates multiple approaches to extract text from PDF documents:

• PyPDFLoader: Fast, basic PDF text extraction
• PDFMinerLoader: Detailed layout-aware extraction
• TesseractBlobParser: OCR-based extraction for scanned PDFs with image content

Features shown:

• Loading PDF files
• Extracting text and metadata
• Handling PDFs with images
• Processing documents for downstream RAG tasks

Core Dependencies

Package	Purpose
langchain	LLM orchestration framework
langchain-community	Community integrations (loaders, utilities)
pypdf	PDF parsing and manipulation
pdfminer.six	Advanced PDF text extraction
pdfplumber	PDF data extraction tool
pytesseract / rapidocr-onnxruntime	Optical Character Recognition (OCR)
openai	OpenAI API integration
beautifulsoup4	HTML/XML parsing
python-dotenv	Environment variable management

Usage

Running Notebooks

Start Jupyter and explore the examples:

jupyter notebook

Navigate to 1_document_loaders/pdf_loader.ipynb to see:

• PDF loading techniques
• Text extraction methods
• Image extraction and OCR processing
• Document metadata handling

Python Script

Run the main application:

python main.py

Next Steps / Roadmap

This project covers the first critical step of RAG: Document Loading. Typical RAG pipeline progression:

1. ✅ Document Loading - Extract text from various sources
2. 📋 Document Chunking - Split large documents into manageable pieces
3. 🔤 Embeddings - Convert text to vector representations
4. 🗄️ Vector Store - Store and index embeddings for retrieval
5. 🔍 Retrieval - Find relevant documents based on queries
6. 🤖 Generation - Use LLMs to synthesize responses from retrieved context

Example Workflow

from langchain_community.document_loaders.pdf import PyPDFLoader

# Load PDF document
loader = PyPDFLoader("document.pdf")
documents = loader.load()

# Process documents (next steps)
# - Split into chunks
# - Generate embeddings
# - Store in vector database
# - Retrieve relevant chunks for queries
# - Generate answers with LLM context

Configuration

Create a .env file for sensitive configuration:

OPENAI_API_KEY=sk-your-key-here
PDF_PATH=path/to/documents

Load in your code:

from dotenv import load_dotenv
import os

load_dotenv()
api_key = os.getenv('OPENAI_API_KEY')

Troubleshooting

PDF Loading Issues:

• Ensure PDF file paths are correct and file exists
• For image-heavy PDFs, use extract_images=True with TesseractBlobParser
• Check that pytesseract is properly configured (see OCR Setup above)

Import Errors:

• Use correct import paths: from langchain_community.document_loaders import PDFMinerLoader
• Verify all dependencies are installed: pip list

Memory Issues:

• For large PDFs, process in chunks or use streaming loaders
• Consider batch processing multiple documents

Resources

• LangChain Documentation
• RAG Papers & Research
• OpenAI API Reference

License

This project is open source and available for educational and development purposes.

Contributing

Contributions are welcome! Please feel free to:

• Add more document loader examples
• Implement vector stores and retrieval
• Add LLM integration examples
• Improve documentation

---

Happy learning! 🚀 Explore the notebooks to see RAG in action.